The notion of markedness has played an important role in linguistic theorizing for decades. Originating in Prague School phonological theory, it has also found an important place in morphology, syntax, and semantics (Andersen 1989). In general terms, markedness theory claims that when alternative forms are available, such as alternative construction types or paraphrases, one member of the set, the unmarked member, is more prototypical, is more frequent, has a wider distribution, etc. Within the psycholinguistic literature, it has been argued that the unmarked member of a set will typically be both cognitively and structurally less complex, and hence easier to process than its marked counterpart, a factor which also suggests that, all things being equal, the unmarked member should be acquired earlier than the marked member (e.g., Slobin 1973).

Most research on Algonquian languages, of which Cree is a typical example, is based on collections of narrative texts (Wolfart 1973; Dahlstrom 1986; James 1986). Although there is nothing intrinsically wrong with this approach, the use of one particular type of database in such an extensive amount of research lends itself to a genre-biased description of the language. In oral cultures, many narrative texts are typically preplanned (Chafe 1985) and therefore will have, according to researchers in discourse analysis, many of the features of preplanned texts such as complete and longer sentences, higher clause density and a larger proportion of subordinate clauses (Brown and Yule 1985:151–117; Biber 1988:47). In addition, other language-specific features may occur.